Three shortcuts, one default folder
Cmd+Shift+3 for the full screen, Cmd+Shift+4 to drag a selection, Cmd+Shift+5 for a toolbar with recording options — that's the entire built-in screenshot toolkit on macOS, no extra app required. By default, every capture lands on the Desktop as a PNG named "Screenshot" plus the date and time, and a floating thumbnail pops up in the bottom-right corner so you can mark it up, share it, or drag it away on the spot.
Changing where they save
If you don't want the Desktop buried in screenshots, the "Options" menu inside the Cmd+Shift+5 toolbar lets you redirect the default save location to Documents, the clipboard, or any folder you pick. If you're comfortable with Terminal, a single defaults write com.apple.screencapture location command does the same, followed by a quick restart of the Dock process to make it stick. Either way, you've only changed which pile they land in — how to organize or find them later is still an open question.
The trouble starts after they're saved
Wherever they land, the filename macOS gives them is just a timestamp — "Screenshot 2026-07-20 at 11.32.15 AM.png." That's fine for three or four shots; once you've got dozens or hundreds, filenames and thumbnails alone won't tell you which one was that error message and which was that receipt. Spotlight can match filenames, but a timestamp carries no information worth searching for — and Spotlight doesn't read the text printed inside the image, so "search for a word that appeared in a screenshot" simply isn't something the built-in tools do.
A different approach: not where, but whether you'll remember it
Built-in macOS Screenshots vs. a Screenshot Tool That Organizes Itself weighs the full tradeoff; here's just what Shotbox does at this step: using the same hotkey, it reads the text inside each shot on your Mac, gives it a title, and drops it straight into a searchable index — no need to track which folder it's in. Recognition runs on Apple's own Vision framework, entirely on-device, and works offline. Worth being upfront about: OCR handles clean printed text well, but handwriting, heavily rotated shots, or dense multi-column layouts (a whole PDF page crammed into one screenshot, say) will drag accuracy down — that's a limit of on-device OCR itself, not something a cloud service would automatically fix. For exactly what permissions and network behavior are involved, see does a screenshot tool upload what's on your screen.
